Retrieval-based Question Answering on ChatDoctor
13.6ROUGE-LOrigin
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| OriginModel=bge-large-en-v1.52026.03 | 13.6 | 16.2 | 100 | 100 | 100 | — | |
| OriginModel=all-MiniLM-L6-v22026.03 | 13.5 | 16 | 100 | 100 | 100 | — | |
| OriginModel=Qwen3-0.6B2026.03 | 13.2 | 15.6 | 100 | 100 | 100 | — | |
| Multi-Agent Semantic Rewriting FrameworkBackbone LLM=GPT-4o-mini2026.06 | 10.4 | — | — | — | — | 12.2 | |
| KG-PrivRAGBackbone LLM=GPT-4o-mini2026.06 | 10 | — | — | — | — | 11.9 | |
| Naive RAGBackbone LLM=GPT-4o-mini2026.06 | 9.5 | — | — | — | — | 9.3 | |
| Single AgentBackbone LLM=GPT-4o-mini2026.06 | 9.2 | — | — | — | — | 10.8 | |
| SAGEBackbone LLM=GPT-4o-mini2026.06 | 9.1 | — | — | — | — | 11.7 | |
| Adv-AnonBackbone LLM=GPT-4o-mini2026.06 | 8.9 | — | — | — | — | 10.3 | |
| TRIP-RAGModel=bge-large-en-v1.52026.03 | 8.8 | 8.6 | 67.2 | 75.2 | 77.8 | — | |
| 0-shotBackbone LLM=GPT-4o-mini2026.06 | 8.8 | — | — | — | — | 8.7 | |
| TRIP-RAGModel=Qwen3-0.6B2026.03 | 8.4 | 8.3 | 72.1 | 78.4 | 81 | — | |
| TRIP-RAGModel=all-MiniLM-L6-v22026.03 | 8.1 | 7.7 | 70.3 | 76 | 78.9 | — | |
| RedactModel=Qwen3-0.6B2026.03 | 7.8 | 7.8 | 43 | 52.4 | 56.8 | — | |
| LPRAGModel=Qwen3-0.6B2026.03 | 7.5 | 7 | 38.9 | 47.7 | 51.3 | — | |
| RedactModel=bge-large-en-v1.52026.03 | 7.5 | 7.2 | 41.5 | 51.4 | 54.7 | — | |
| RedactModel=all-MiniLM-L6-v22026.03 | 7.4 | 6.7 | 41.6 | 51.2 | 54.5 | — | |
| AttrPromptBackbone LLM=GPT-4o-mini2026.06 | 7.3 | — | — | — | — | 8 | |
| LPRAGModel=bge-large-en-v1.52026.03 | 7.2 | 6.6 | 34.6 | 44.6 | 47.8 | — | |
| ParaphraseModel=bge-large-en-v1.52026.03 | 6.7 | 6.3 | 13 | 18.8 | 21.1 | — | |
| ParaphraseModel=Qwen3-0.6B2026.03 | 6.6 | 6 | 16.1 | 23.5 | 26.6 | — | |
| ParaphraseModel=all-MiniLM-L6-v22026.03 | 6.6 | 5.9 | 10.5 | 18 | 20.5 | — | |
| LPRAGModel=all-MiniLM-L6-v22026.03 | 6.5 | 5.7 | 36 | 44.3 | 48 | — |